SPECIAL SECTION ON FUTURE NETWORKS: ARCHITECTURES, PROTOCOLS, AND APPLICATIONS

Received January 29, 2017, accepted February 19, 2017, date of publication March 2, 2017, date of current version May 17, 2017.
Digital Object Identifier 10.1109/ACCESS.2017.2676004

A Balanced Energy-Consuming and

Hole-Alleviating Algorithm for Wireless
Sensor Networks
NAEEM JAN1 , NADEEM JAVAID1 , (Senior Member, IEEE), QAISAR JAVAID2 , NABIL ALRAJEH3 ,
MASOOM ALAM1 , ZAHOOR ALI KHAN4 , (Senior Member, IEEE), AND IFTIKHAR AZIM NIAZ1
1 COMSATS Institute of Information Technology, Islamabad 44000, Pakistan
2 International
Islamic University, Islamabad 44000, Pakistan
3 CAMS, Department of Biomedical Technology, KSU, Riyadh 11633, Saudi Arabia
4 Computer Information Science, Higher Colleges of Technology, Fujairah 4114, UAE

Corresponding author: N. Javaid (nadeemjavaidqau@gmail.com)

This work was supported by King Saud University through International Scientific Partnership Program under Grant 0053.

ABSTRACT In wireless sensor networks, energy balancing and energy efficiency are the key requirements
to prolong the network lifetime. In this paper, we investigate the problem of energy hole, in which sensor
nodes located near the sink or in some other parts of the network die early due to unbalanced load
distribution. Moreover, there is a dire need to utilize the energy resource efficiently. For this purpose,
balanced energy consumption and hole alleviation, and energy-aware balanced energy-consuming and hole-
alleviating algorithms are proposed. These algorithms balance the distribution of load along with efficient
energy consumption. An optimal distance and energy-based transmission strategy with least expected error
rate is adopted to forward the data packets of different sizes. Furthermore, the data distribution between high-
energy consuming nodes and low-energy consuming nodes in each corona is analyzed. This distribution
enables the proposed algorithms to outperform their counterparts in term of network lifetime, balanced
energy consumption, and throughput on the cost of increased end-to-end delay.

INDEX TERMS Wireless sensor networks, energy balancing, energy hole alleviation, throughput
maximization, linear optimization.

I. INTRODUCTION with substantial amount of unused aggregated energy. Hence,

Wireless Sensor Networks (WSNs) consist of multiple sen-
sor nodes which sense the parameters of interest and report
to the sink(s). These networks have the diverse range of
applications because of their data gathering capability in
hostile, and remote areas [1]– [3]. Generally, deployed nodes
are equipped with limited energy; hence, efficient energy
consumption is of prime importance. [4], [5]. Sink(s) is the
central entity of the network; hence, it is generally located
at some central position. The data generated in network is
forwarded towards the sink with multi-hop communication
paradigm. The nodes nearer to the sink forward the aggre-
gated data (data generated plus data received); hence, energy
consumption of these nodes is relatively higher than the
farther nodes. Moreover, due to random deployment there
may occur the void holes in the network [6]. The presence of
void hole further enhances the uneven energy consumption
and leads towards the creation of energy holes. The void and
energy holes disjoin the networks into disconnected regions,
with the balanced energy consumption not only the problem
of energy holes is reduced, but also the lifetime of the whole
network is increased [7]– [10].
The sensor nodes report the sensed data to sink either
in direct or in multi-hop transmission manner. Although,
direct transmission is an easy way for reporting the data;
however, if all nodes transmit data directly then the energy
of farther nodes deplete much quickly. Similarly, in multi-
hop transmission manner the farther nodes send data to its
internal corona nodes. The data packets which are transmitted
by the outer corona nodes is mainly distributed in the first
radius of the sink. Therefore, the probability of energy hole is
maximum near the sink. It is because that smaller is the trans-
mission range more data is forwarded by the sensor nodes
located near the sink. However, when the transmission range
increases, most of the data is transmitted directly to the sink.
Therefore, the load on the first corona nodes decreases and
increases on the outer corona nodes. Therefore, it can be

2169-3536
2017 IEEE. Translations and content mining are permitted for academic research only.
6134 Personal use is also permitted, but republication/redistribution requires IEEE permission. VOLUME 5, 2017
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
N. Jan et al.: BECHA Algorithm for WSNs
concluded that energy consumption is mainly dependent on
the data size and the distance parameter [11].
Recently, the optimization techniques efficiently improve
the network lifetime by balancing the energy consumption.
In which the adaptive range adjustment strategy efficiently
enhances the network lifetime [12]. An efficient transmission
strategy plays an important role in maximizing the lifetime
of a network, which is dependent on the life of sensor nodes.
To maximize the network lifetime it is necessary to minimize
the death rate of sensor nodes. The death of the sensor nodes is
mainly caused by the exhaustion of energy due to unbalanced
load distribution. The premature death of sensor nodes leads
to serious energy and coverage holes problem. To avoid
the energy hole problem around the sink, a wireless sensor
network energy hole alleviating (WSNEHA) algorithm is
proposed in [13]. The energy consumption of first radius sen-
sor nodes is balanced by the algorithm. However, the energy
consumption of sensor nodes located in the other regions is
also very uneven. Moreover, there is dire need to utilize the
energy resource efficiently. In order to achieve high through-
put while being energy efficient, it is necessary to acquire
an optimal distance which is free of location error. For this
purpose, an energy condition mean square error algorithm
is proposed in [17], which considers a distance-based metric
to enhance the throughput. Although the algorithm achieves
high throughput, though energy hole problem is emerged due
to the iterative selection of sensor nodes.
Different algorithms are proposed in the literature and
their design either focuses on balanced energy consumption
or on high throughput. We consider both the aspects to
enhance the throughput with energy efficiency. We propose
a balanced energy consuming and hole alleviating (BECHA)
algorithm [18] to level the load distribution of entire network.
Our first contribution is to extend the work of WSNEHA
algorithm by balancing the load of entire network. In second
contribution i.e., an energy-aware balanced energy consum-
ing and hole alleviating algorithm (EA-BECHA), our aim is
to enhance the throughput by minimizing the packet drop
ratio and prolong the network lifetime through efficiently
utilizing the energy resource.
Rest of the paper is organized as follows. Section II rep-
resents a brief summary of literature review. Section III
describes the system model, while the proposed algorithms
are presented in section IV. Linear programming based math-
ematical modeling of simulation parameter are proposed in
section V. The performance evaluation is given in section VI,
and performance trade-off in section VII. Finally, the sum-
mary of contributions along with future work is concluded
in section VIII.
II. RELATED WORK
In order to mitigate the uneven energy consumption different
approaches are adopted not only to balance the overall energy
dissipation but also to increase the network lifetime and
throughput. The routing protocols are summarized in Table 1.
A distributed coverage hole repair algorithm (HORA) for
VOLUME 5, 2017
WSNs is proposed by Sahoo and Liao [14], which employs
the node mobility of sensor nodes to reduce the energy hole
problem. To maximize the coverage area by minimizing the
overlapping region each sensor node checks its status of either
cross triangle (CT), hidden cross triangle (HCT), and non-
cross triangle (NCT) node. A mobile node is selected on
the bases of higher degree of overlapping region and by the
condition that its existing connectivity and coverage may not
be disturbed. Therefore, the neighbor nodes with higher over-
lapping region will move first to repair the hole. The algo-
rithm ensures no more coverage hole creation, outperforms
in network lifetime and achieves more throughput. However,
HORA is not suitable for delay sensitive applications due to
the hole repair process.
Latif et al. [15], propose a spherical hole repair tech-
nique (SHORT) to repair the coverage and energy holes prob-
lem in underwater wireless sensor networks (UWSNs). The
proposed technique is composed of three phases; knowledge
sharing phase (KSP), network operation phase (NOP), and
hole repair phase (HRP). Similarly, the nodes with higher
degree of overlapping region are responsible for repair of
energy and coverage holes. The results show good improve-
ment in term of energy consumption and network lifetime
with trade-off of high end-to-end delay.
An optimal distance-based transmission strategy (ODTS)
using ant colony optimization (ACO) is proposed in [16].
The algorithm introduces two notions i.e., most energy
efficient distance (MEED), and most energy balance dis-
tance (MEBD). The local optimal transmission distance
is determined by MEED, while the global optimal trans-
mission distance is achieved by balancing the load of all
nodes through MEBD. Although, ODTS balances the energy
consumption of entire network and efficiently utilizes the
energy resource. Therefore, network lifetime maximization is
achieved through energy balancing. However, the proposed
algorithm is inefficient for random and sparse deployment;
moreover, the energy hole problem is emerged due to the
iterative selection of nodes.
Chen et al. [19], propose a power saving mobicast rout-
ing protocol for UWSNs to mitigate the energy hole prob-
lem, which is caused by the ocean current and non-uniform
deployment of sensor nodes. To overcome the energy hole
problem 3-D zone of relevance (3-D ZORt ), 3-D zone of
forwarding (3-D ZOFt+1 ) and an apple peel technique is
introduced. The energy hole problem is eliminated through
expending the adaptive segments, while the energy effi-
ciency is achieved by the collection of data through mobile
autonomous underwater vehicles (AUVs). The algorithm
efficiently maximizes the coverage area by covering the
hole region, and also maximizes the packet delivery ratio.
However, if the location of nodes is not predetermined then
the algorithm will not perform well. Moreover the algorithm
also pay cost due to message overhead.
A geographic and opportunistic routing protocol (GEDAR)
based on depth-adjustment control strategy for recovery of
void nodes is proposed by Countinho et al. [20]. GEDAR
6135

uses greedy forwarding strategy for finding the neighbors
and adjust the depth of void nodes either by buoyancy-
based or winch-based. Although, the algorithm outperforms
in packet delivery ratio, and less number of re-transmissions.
However, the geographic and opportunistic routing protocols
are resilient to location error and leads to high end-to-end
latency. Furthermore, the algorithm utilizes high energy due
to multiple data packets sending to neighbors, and also due to
void node recovery procedure.
To avoid the energy hole problem in UWSNs, the authors
propose balanced routing (BR) scheme in [21]. To miti-
gate the energy hole problem around the sink, data of the
1st corona node is allocated to the 2nd corona node using
two-hop transmission range 2r. Although, the algorithm
alleviates the energy hole problem in the first corona;
however, the corona two which is the two-hop neigh-
bor of sink treats as the one-hop neighbor of sink,
which leads to load imbalance and creates energy hole
problem.
In order to alleviate the energy hole problem, a mobile
sink-based adaptive immune energy-efficient clustering pro-
tocol (MSIEEP) is proposed in [22]. The adaptive immune
algorithm (AIA) based on energy dissipated is used to find
the favorable number of cluster heads (CHs); moreover, AIA
is also used to supervise the mobile sink about their sojourn
location. Although, the algorithm efficiently utilizes the
energy resource, prolongs the network lifetime with increased
throughput. However, still the energy hole problem arises due
to unbalanced load distribution.
Tunca et al. [23] present a virtual ring routing structure
with mobile-sink approach to reduce the hotspot problem.
To collect the data from nodes, mobile-sink moves in a pre-
defined path. The recent sojourn location of sink is deter-
mined by the high-tier sensor nodes, whereas low-tier sensor
nodes communicate with the high-tier nodes to take the recent
location information of sink. Moreover, the anchor nodes
which act as relay nodes to sink are also introduced to suc-
cessfully forward the data packet to sink. The algorithm max-
imizes the network lifetime by utilizing the energy resource
efficiently and also reduces the end-to-end delay. However,
due to erroneous location information the algorithm results
in low packet delivery ratio.
A weighted rendezvous planning (WRP) algorithm based
on rendezvous points (RPs) is presented in [24] to reduce the
traveling path and energy hole problem. A hybrid moving
strategy using mobile-sink is adopted to collect the aggre-
gated data from the RPs. The nodes advance their data to the
nearest RPs using multi-hop transmission strategy. Similarly,
weights are assigned to each node based on the hop-distance
and number of packets in buffer. The selection of RP is
conditioned to the weight factor i.e., the node with maximum
weight factor is selected as RP. Although, network lifetime is
achieved by minimizing the energy consumption. However,
the algorithm performs inefficiently for data gathering in void
regions. Furthermore, due to location error less throughput is
achieved.
6136
N. Jan et al.: BECHA Algorithm for WSNs
Lin et al. [25], Shamsan et al. [26], and Liao et al. [27]
introduce clustering and mobile agents in WSNs for energy
balancing. A virtual tree based cellular topology is presented
to balance the uneven load distribution. In order to alleviate
the energy hole problem and establish a reliable path for data
delivery, the mobile nodes are moved to the hotspot region
to collect the sensed data. Moreover, time slot is defined to
avoid congestion in communication. The algorithm achieves
network lifetime by balancing the load with the cost of high
delay.
A general self organized tree-based energy-balance rout-
ing protocol (GSTEB) is introduced by the author in [28].
In order to balance the load distribution and minimize the
energy consumption a three step technique is proposed,
namely tree constructing phase, data collection and transmis-
sion phase, and information exchange phase. A root node
is selected on the basis of high residual energy between the
neighbors or itself, which is responsible for the collection and
transmission of data packets. Network lifetime is achieved
through efficient energy balancing, though with the cost of
delay and message overhead.
An analytical model is presented in [29] which investi-
gate two approaches i.e., hierarchical deployment, and data
compression and aggregation. In order to balance the load
distribution and mitigate the energy hole, a mobile-sink along
with assisting nodes are introduced. The mobile sink collects
the aggregated data from the assisting nodes which are at the
top layer of normal node and responsible for the transmission
of these leaf node data. Energy balancing is achieved with
which energy hole problem is also reduced. However, the
algorithm is inefficient for delay sensitive application. Simi-
larly, an analytical model is proposed in [30] to evaluate the
energy consumption, and network lifetime. The observations
are considered during the entire network lifetime and also
under given percentage of dead nodes. Moreover, with the
proposed model spatio-temporal issues of emerging energy
hole are investigated. The emerging time of death appear-
ance and boundary of the energy hole is also determined.
The analytical results of the scheme indicate efficient energy
utilization and network lifetime longevity.
The use of mobile relay nodes is introduced by the author
in [31] to remove the energy hole problem and to solve the
issue of localization of sink. For this purpose, the entire
network is divided into different clusters and mobile relay
nodes are applied in the network to collect data from sensor
nodes and transmit further to the sink. The relay nodes are
considered as the mobile sink of sub-network regions. The
algorithm is suitable for event-driven and continuous data
delivery schemes. Moreover, with the help of mobile relay
nodes energy efficiency, load balancing, and network lifetime
maximization are achieved. However, the network become
disjoint if the relay nodes are inefficient to transmit the data
packets.
A two-fold game theoretic routing protocol, namely evolu-
tionary game theory (EGT) and classical game theory (CGT)
is proposed in [32] to balance the energy consumption of
VOLUME 5, 2017
N. Jan et al.: BECHA Algorithm for WSNs

TABLE 1. Comparison of routing protocols.

VOLUME 5, 2017 6137


TABLE 1. (Continued.) Comparison of routing protocols.
entire network. The former takes the responsibility of energy
balancing of different regions, while the latter is used at
the nodes level balancing. The task of region level energy
balance (RLEB) is to forward a uniform proportion of data
packet to all sub-regions. Moreover, the algorithm defines
a fitness function to recognize the switching probability
between sub-regions which is expressed in term of gain and
cost of consumption for advancement. While the node level
energy balance (NLEB) is used to select one forwarder from
NK sensor nodes in the sub-region to advance the packet.
The algorithm shows good improvement in network lifetime,
packet delivery ratio, and average energy consumption. How-
ever, the geographic routing protocols are resilient to location
error and the sensor nodes require positioning hardware.
In [33], the authors present a routing strategy to divert
the data packet through specific angle around the energy
hole polygon. The algorithm determines the area and angle
of approximate polygon when an energy hole is emerged,
whereas the range of polygon is determined by αmin and αmax .
By indicating the hotspot region a hole core informa-
tion (HCI) message is dispatched to its neighbors. To transmit
the suspended data packet, a new route is determined and
the useful information is disseminated through the diverted
route. Despite the fact that the algorithm outperforms in
throughput, though results in long end-to-end delay, more
energy depletion and longer routing path.
A synchronization of nodes in adjacent annulus (SNNA) is
introduced in [34] to balance the energy usage and mitigate
the energy hole problem. In order to minimize the energy
hole problem around the sink. The sensor nodes are deployed
non-uniformly such as from low density to high density
near the sink. Moreover, a synchronization based strategy
is put forwarded to transmit and receive the data between
adjacent annulus nodes. Each node in the network checks
the residual energy and distance to find the optimal parent
in the inner annuls. Furthermore, a sleep-mode scheduling
mechanism is also introduced to minimize the energy con-
sumption of entire network. Although, the algorithm max-
imizes the network lifetime. However, with trade-off of
long end-to-end delay, and packet loss due to sleep-mode
scheduling.
6138
N. Jan et al.: BECHA Algorithm for WSNs
In [35], the authors present cluster head and relay selec-
tion mechanism to resist the energy hole problem caused
by the first gradient sensor nodes. The procedure is per-
formed in three phases i.e., identify the node that requires
a relay, relay selection mechanism, and taking advan-
tage of hybrid clustering and routing (HCR). Moreover,
a cost function based on energy threshold is defined by
the algorithm for selection of relay nodes. Results indi-
cate good energy balancing and network lifetime longevity
with the cost of high transmission latency and end-to-end
delay.
In order to mitigate the energy sink-hole problem,
a three-tier architecture composed of immobile sink, immo-
bile sensor nodes, and mobile proxy sink is presented by
the author in [36]. To minimize the imbalance load on sink
neighbors, the sensor nodes deployed in the network are
attached with heterogeneous initial energy, such that more
energy is allocated to the sink proximal neighbors. Moreover,
multiple mobile proxy sinks are responsible for the collection
and distribution of data from sink neighbors to immobile sink.
The algorithm takes advantage of mobile sink and alleviate
the energy hole problem efficiently around the sink neigh-
bors. However, the nodes located in a hop far away from the
sink still facing the problem of unbalanced load distribution
and energy hole problem.
A probabilistic coverage hole detection algorithm (HDRE)
is proposed in [37] to maximize the network lifetime by
avoiding energy hole. Sensor nodes with low residual energy
die very early; consequently, yields to the emergence of
more energy hole. To cope the limitation the algorithm takes
decision on the probabilistic analysis and selects an optimal
node to repair the coverage hole on the basis of residual
energy. HDRE yields no more coverage hole and results in
network lifetime longevity with the cost of repair and energy
consumption.
The authors in [38] propose a relay node replacement strat-
egy to eliminate the energy hole problem. The relay nodes are
uniformly distributed in specific area with equidistant from
the sink. The sensor nodes send their data to the relay nodes
when their energy is less than a specific threshold. The relay
nodes are responsible for data forwarding to the sink. Due to
VOLUME 5, 2017
N. Jan et al.: BECHA Algorithm for WSNs

FIGURE 1. Network model of WSNEHA algorithm. FIGURE 2. Network model of the BECHA algorithm.

the selection of relay nodes network lifetime longevity and

energy balancing is achieved. where ETx shows the overall energy consumed in data trans-
mission. d is the distance between source node and neighbor
III. SYSTEM MODEL nodes or sink. Cbase , α, and F are the radio channel constants,
We consider a WSN, similar to [6], [16] with network radius R s is the packet size, < is the information rate. Cefs and Camp
and transmission radius r. The network is further divided into are the amplifier energy which depends on distance parameter
N concentric logical circles denoted as C1 , C2 , · · · , CN , each and bit error rate. If α = 2 then Cbase = Cefs and if α =
q4 then
with thickness t = NR , as shown in Fig. 1. C
efs
Cbase = Camp . do is the reference distance (do = Camp ),
A. NODE ATTRIBUTES respectively.
Sensor nodes with density σ are deployed uniform randomly ( b0
[ Cefs ( 2 b0−1) d 2 + bF0 ] s if d ≤ do
around a sink. The sink is free of energy constraint, while ETx (bo , d) = ( 2b0 −1) 4
(2)
the sensor nodes are equipped with limited battery power [ Camp b0 d + b0 ] s if d > do
F

source. Energy is majorly used in data transmission which F

is a precious and limited resource. Moreover, sensor nodes ERx = s, (3)
bo
are stationary and have the same initial energy and maximum
transmission range. where, ERx is the energy consumed on data reception.

B. NETWORK TOPOLOGY IV. PROPOSED ALGORITHMS

Let K = {k1 , k2 , · · · , kn } is the set of sensor nodes, and
L = {l1 , l2 , · · · , lm } be the one-hop neighbors set of K . Let
graph G(t)=(V , E(t)) with |V | = K represents the graph at
time t, where V is the set of vertices and E is the set of links
at time t. Two sensor nodes i and j are neighbors when they
are directly connected through an edge E ∈ {eij }. The data
transmission is successful if every sensor node i is connected
through a communication channel {eij } with senor node j
onward to sink. The flag ∧ = 0 indicates that the information
is not disseminated to its neighbors.
C. ENERGY CONSUMPTION MODEL
The energy consumption model of receiving and sending bo
bits of data at a distance d is given as [11]:
s We consider a network model where nodes are deployed
ETx (bo , d) = <[ C( 2 τ < − 1) + F] τ,
( 2b0 − 1) α F nodes gather the information from environment and forward
= [ Cbase d + ] s, (1)
b0 b0
Energy balancing is in direct relation with the network life-
time. The network lifetime can be prolonged efficiently by
implementing good balancing strategy. For this purpose, we
propose BECHA and EA-BECHA algorithms.
A. THE BECHA
Initially the sensor nodes are deployed, which are unaware of
network information. After the network configuration each
node is informed with a hello message which contains the
information about sink and neighbors attributes i.e., Node ID,
status of link, received signal strength and residual energy of
neighbor nodes.
1) NETWORK MODEL
uniform randomly around a sink, as shown in Fig. 2. The
either to the next hop forwarding nodes or sink directly.

VOLUME 5, 2017 6139

 N. Jan et al.: BECHA Algorithm for WSNs

The data is forwarded to the neighbor nodes according to the the distance between the transmitter and receiver is minimum.
energy level and distance parameter. The BECHA algorithm balances the energy consumption by
allocating and forwarding the data packets of different sizes
2) LOAD DISTRIBUTION to each corona sensor nodes except the outermost corona.
The sensor nodes in the outermost corona send data firstly In multi-hop communication, it is obvious that there exist
to the internal neighbors. Similarly, the internal corona minimum energy consuming nodes and maximum energy
nodes are responsible for the transmission of aggregated consuming nodes. For this purpose data packet of different
data. The process iterates until the sink successfully receives sizes is distributed according to the energy level of ith − 1
the data packet. Therefore, more energy is consumed by corona nodes. Assume that a transmitting node t1 sends data
the inner coronas’ sensor nodes. Moreover, in each corona to a receiving node r1 located at a distance S1 from t1 .
there exists maximum energy consuming and minimum It can be concluded that more amount of data is transmit-
energy consuming nodes. The BECHA algorithm is pro- ted if the destination changes to the receiver r2 , such that
posed to balance the energy consumption by distributing r2 < r1 .
the load uniformly. The sensor node which participates (
more in the data transmission consumes energy quickly. The (1Dxt (Cefs (2bo − 1)d 2 + 2F) − F)/bo if d ≤ do
1E =
data packets of different sizes 1D is calculated in [13] is (1Dxt (Cefs (2bo − 1)d 4 + 2F) − F)/bo otherwise
simplified as: (5)
(
1Ebo + F/(Cefs (2bo − 1)d 2 + 2F) if d ≤ do where, 1E is the amount of energy which is consumed
1Dt =
x
1Ebo + F/(Camp (2bo − 1)d 4 + 2F), otherwise according to the data size and distance parameter.
(4)
B. THE EA-BECHA
where 1Dxt is the data packet which is allocated according
The goal of EA-BECHA is to alleviate the energy hole
to the energy level of neighbor nodes. Similarly, 1E is the
problem by minimizing the energy consumption of those
corresponding energy that is consumed on forwarding this
sensor nodes which are selected iteratively in each round.
data.
Consequently, the energy of nodes exhausts very early and
Algorithm 1 BECHA causes energy hole problem. The analysis indicates that there
1: procedure Initialize all parameters
exits maximum and minimum energy consuming nodes in
2: Ei :={E1 , E2 , E3 , · · · , En }; each corona. Therefore, the energy hole problem is elim-
3: SetNeigh := {a1 , a2 , a3 , · · · , an }; inated by allocating the data of maximums to minimum
4: SetNeighDist := {d11 , d12 , d13 , · · · , dnn }; energy consuming nodes. The task is performed by select-
5: GetInfo (SetNeig, Ei , SetNeighDist) ing an optimal forwarder based on high residual energy
6: for each node i ∈ N do and minimum distance from the sink. The EA-BECHA
7: generate Rnd calculates the forwarding factor FF of neighbor nodes,
8: Find MaxEi r, MaxEi+1 r; such as:
9: Find MinEi r, MinEi+1 r; Nje
10: for di < di+1 &&MaxEi r < MaxEi+1 r do FF ∝ , ∀ j = 1, 2, 3, · · · , m, (6)
d(i,j)
11: if MaxEi > Th then
12: Select forwarder; where, Nje shows the energy of neighbor nodes j, and d(i,j)
13: Update SetForwd is the corresponding distance between ith and ith − 1 corona
14: Eavg : Eavg − 4Ei+1 nodes. The FF of all neighbor nodes is calculated according
15: Update SetEi , MaxEi r, MaxEi+1 r to Eq. 6. Where, the forwarder selected Fs in Eq. 7 shows
16: else that the neighbor node with high FF is selected as the next
17: break; hop forwarder.
18: end if
19: end for Fs = argmax(FF ). (7)
20: end for
Similarly, in each corona the best forwarder is selected on the
21: end procedure
basis of high FF . It is obvious that the energy consumption is
high when the distance between the transmitter and receiver
is maximum. Therefore, the optimal forwarder is selected
3) ENERGY CONSUMPTION on the bases of high residual energy and minimum distance
The energy consumption of sensor nodes is in direct relation from the sink. A routing path at a distance D between a
with the distance and data size. More is the transmission sender and receiver consists of n hops, and n − 1 intervening
distance and packet size more is the energy consumed by optimal forwarder nodes [16]. The overall energy dissipa-
sensor nodes. While less amount of energy is consumed when tion is minimum when each node transmits the data through

6140 VOLUME 5, 2017

N. Jan et al.: BECHA Algorithm for WSNs

optimal distance d = D/n. The total rate of dissipation is 1) DATA DISTRIBUTION

given as: The aggregated data is transmitted by the sensor nodes in an
efficient manner to achieve both high energy efficiency and
Etot = (n)ETx + (n − 1)ERx good energy balancing. The algorithm allocates data packets
D 2bo − 1 α F D F of different sizes according to the FF of neighbors and select
= [Cefs d + ] + ( − 1) ,
d bo bo d bo the neighbor with high FF . Similarly, the process iterates
D 2bo α 2D F for all coronas maximums and minimums energy consuming
= Cefs d +( − 1) , nodes.
d bo d bo
2DF F 2bo − 1 α 2) ENERGY CONSUMPTION
= − + DCefs d ,
bo d bo bo The energy consumption of each sensor node is dependent on
2bo − 1 α−2 2F data size and distance between the transmitter and receiver.
×(α − 1)Cefs Dd − Dd −2 = 0. (8)
bo bo In EA-BECHA the energy consumption is efficiently reduced
through the optimal forwarding strategy. The optimal for-
By taking the derivative of Eq. 8 equal to 0, the energy optimal
warder is selected on the bases of optimal distance and high
distance EOD can be deduced,
energy level.
s
2F V. LINEAR OPTIMIZATION
dEOD = α . (9)
(α − 1)2bo−1 Cefs Linear optimization is widely used mathematical technique to
find the optimal solution. In linear programming the objective
function is followed by the set of constraints which formulate
Algorithm 2 EA-BECHA the solution [39]. The following parameters are mentioned to
1: procedure Initialize all parameters locate the optimal solution.
2: Ei :={E1 , E2 , E3 , · · · , En };
A. ENERGY CONSUMPTION
3: SetNeigh := {a1 , a2 , a3 , · · · , an };
4: SetNeighDist := {d11 , d12 , d13 , · · · , dnn }; During the transmission phase of EA-BECHA, the data pack-
5: SetForwd := {f1 , f2 , f3 , · · · , fn } ets are transmitted to the sensor nodes with high FF . The
6: GetInfo (SetNeig, Ei , SetNeighDist, SetForwd) energy consumption on these transmissions is defined as: the
7: for each node i ∈ N do energy which is consumed by transmitting and receiving l bits
8: generate Rnd of data at a distance d. The cost function along with the set
9: Find MaxEi r, MaxEi+1 r; of constraints is defined as:
rmax Xn
10: Find MinEi r, MinEi+1 r; X
Min ECxr ∀r ∈ R ∧ x ∈ N , (10)
11: Find EOD
r=1 x=1
12: Calculate FF ∝ Nje /dEOD ;
such that,
13: for di < di+1 &&MaxEi r < MaxEi+1 r do
14: if MaxEi+1 > MaxEi then C1 Ex ≤ Exini ∀x ∈ N ,
: (11)
15: if E(fi ) > Th then C2 : r
t × ETx r
+ t × ERx ≥ Exmin ∀x ∈ N , (12)
x x
16: if Ei > Eavg then C3 Ex ≤ Ex−1 ∀x ∈ N ,
: (13)
17: if MaxEi+1 r < MaxEi r then
C4 dx ≤ dxmin ∀x ∈ N ,
: (14)
18: Select optimalforwarder; rmax
else
X
19: C5 : Max FFxr ∀r ∈ R ∧ x ∈ N . (15)
20: Select next forwarder; r=1
21: Update SetForwd;
Our main objective in Eq. 10 is to minimize the energy
22: Eavg : Eavg − 4Ei+1 ;
consumption EC of each node x. The summation of EC
23: Update MaxEi r, MaxEi+1 r;
over r, accumulates the overall energy dissipation of x. Con-
24: Update Ei ;
straint in Eq. 11 shows the upper bound of energy i.e., each
25: end if
node is initialized with uniform initial energy. Constraint in
26: end if
Eq. 12 indicates that a minimum level of energy is required
27: else
for successful communication, where t shows the total num-
28: break;
ber of transmissions performed by each node to deliver a
29: end if
data packet. The inequality in Eq. 13 prevents the load dis-
30: end if
proportionality. Constraint in Eq. 14 depicts that a mini-
31: end for
mum transmission range is necessary for a successful com-
32: end for
munication. In order to achieve balanced energy consump-
33: end procedure
tion, it is necessary to select an optimal forwarder based on

VOLUME 5, 2017 6141


FIGURE 3. Energy consumption: Feasible region.
maximum FF . Eq. 15 deals with the selection of FF based on
high residual energy and optimal distance.
1) GRAPHICAL ANALYSIS
Consider a scenario where test points are selected according
to the system of inequalities in Eq. 12 to Eq. 15; Cefs = 50 nJ ,
Camp = 50 nJ , d = 60, Psize = 1024. Now the test points are
represented in Eq. 11 to Eq. 15 as:
1 ≤ ETX ≤ 7.69, (16)
0.5 ≤ ERX ≤ 4.096, (17)
1.5 ≤ ETX + ERX ≤ 12. (18)
The test points provided in Eq. 16 to 18 construct the fea-
sible region at points P1 , P2 , and P3 as depicted in Fig. 3.
Each vertex of feasible region shows an optimal solution.
All points on the closed upper half plane validate the
solution as:
at P1 (7, 0.5) = 7.5 nJ ,
at P2 (11.5, 0.5) = 12 nJ , and
at P3 (7, 5) = 12 nJ .
Therefore, it is proved that all the corner points satisfy the
conditions of feasible region subject to the given constraints.
B. PACKETS DROP
During the transmission phase, the data packets are transmit-
ted from source to destination through a wireless channel.
Packets drop is difference between total number of packets
sent by sensor nodes to the number of packets received by
sink successfully. The packets drop may occur due to low link
quality or due to the death of forwarder nodes. To minimize
the number packets drop PDr , the cost function followed by
the set of constraints is defined as:
rmax X
n feasible region on the closed upper half plane at right side as
X
Min cPDr rx ∀r ∈ R ∧ x ∈ N , (19)
r=1 x=1
where,
 at P3 (1000, 3000) = 4000 pkts.
1 if nodes/forwarders transmit packets to sink
c=
0 if nodes communicate with forwarders,
6142
N. Jan et al.: BECHA Algorithm for WSNs
FIGURE 4. Packets drop: Feasible region.
(20)
such that,
C1 : plink ≤ pgood , (21)
C2 : min
Ex ≤ Ex−1 , (22)
C3 : dx ≤ dx+1 .
max
(23)
Our aim in Eq. 19 is to minimize the overall PDr of entire
network. The summation of PDr accumulates the packets
drop of each node x in r rounds. The flag c in Eq. 20 is
incremented each time when a data packet is handed over
by the forwarder node to the sink. Constraint in Eq. 21
deals with the current status of link between a transmitter
and receiver. The probability of current link will be greater
than a minimum threshold i.e., pgood . If the link quality is
less than pgood then the chances of packets drop increase.
Constraint in Eq. 22 illustrates that a minimum energy level is
required for both transmission and reception, while inequality
in Eq. 23 suggests that the receiver would be in the transmis-
sion range for successful communication.
1) GRAPHICAL ANALYSIS
Consider a scenario, where test points are chosen according
to the system of inequalities in Eq. 21 to Eq. 23, such as
PDr = 1000 and packets reached PR = 2000. Now the test
points are represented as:
0 ≤ PDr ≤ 1000, (24)
1000 ≤ PR ≤ 2000, (25)
1000 ≤ PDr + PR ≤ 3000. (26)
The assumed corner points in Eq. 24 to Eq. 26 construct the
depicted in Fig. 4. Each point validates the solution as:
at P1 (1000, 1000) = 2000 pkts,
at P2 (3000, 1000) = 4000 pkts, and
Therefore, subject to the given constraints and intersection of
corner points, the vertex of solution region is validated.
VOLUME 5, 2017
N. Jan et al.: BECHA Algorithm for WSNs
FIGURE 5. Load distribution: Feasible region.
C. LOAD DISTRIBUTION
In order to minimize the energy hole problem, it is necessary
to balance the load distribution of entire network. Our aim
is to level the traffic load in order to avoid early depletion
of energy. The task is performed by forwarding the data to
the nodes with high residual energy. Therefore, the objective
function to level the load distribution LD is defined as:
rmax X
X n
Min LD rx ∀r ∈ R ∧ x ∈ N , (27)
r=1 x=1
such that,
C1 : dpx ≤ dpth
x ∀x ∈ N , (28)
C2 : Ex ≥ Exmin ∀x ∈ N , (29)
C3 : dx ≤ dxmin ∀x ∈ N . (30)
The objective function is defined in Eq. 27 to level the data
load between each corona maximum and minimum energy
consuming nodes. Constraint in Eq. 28 shows that aggregated
load of each node i.e., the data packet transmitted or received
will be less than a specified threshold. The data packets of
different sizes is transmitted to the sensor nodes, according
to the energy level of forwarder nodes. Constraint in Eq. 29
shows a minimum energy threshold and the data packet only
be transmitted by qualifying the condition. More data is
transmitted only if the energy level is high and the distance
between the transmitter and receiver is minimum.
1) GRAPHICAL ANALYSIS
Consider a scenario where the points are selected according
to the system of inequality in Eq. 28 to Eq. 30 such as
bandwidth allocated to forwarder nodes BF = 2500 KHz,
BNF = 500 KHz. Now the test points are represented as:
500 ≤ BNF ≤ 1000, (31)
1000 ≤ BF ≤ 2500, (32)
1500 ≤ BF + BNF ≤ 3500. (33)
The corner points provided in Eq. 31 to 33 construct the
feasible region at points P1 , P2 , and P3 as depicted in Fig. 5.
VOLUME 5, 2017
Each vertex of feasible region shows an optimal solution. All
points on the closed upper half plane validate the solution as:
at P1 (500, 2500) = 3000 nJ ,
at P2 (1500, 2500) = 4000 nJ , and
at P3 (500, 3500) = 4000 nJ .
Therefore, it is proved that all the corner points satisfy the
conditions of feasible region subject to the given constraints.
D. END TO END DELAY
The time elapsed, during the data transmission from source
to destination is termed as end-to-end delay. The end-to-
end delay is composed of transmission delay, processing
delay, queuing delay and propagation delay. We termed
the transmission delay, processing delay and queuing delay
as nodal delay, while consider the queuing delay as zero.
In multi-hop transmission the end-to-end delay is high due
to more processing time; moreover, it also increases with the
increase in distance and density. Therefore, the cost function
for minimizing the end-to-end delay E2E is given as follows:
Min (c E2E), (34)
where,
(
ND (i, j) + ND (j, s) + PD (i, j) + PD (j, s) if j 6= s
E2E =
ND (i, s) + PD (i, s) if j = s
(35)
such that,
C1 : ND ≤ NDmin , (36)
D .
C2 : PD ≤ Pmin (37)
The cost function is defined in Eq. 34 to minimize the overall
E2E delay, where ND and PD in Eq. 35 show the nodal delay
and propagation delay. Similarly, the ND (i, j) is the delay
occurred when the sensor nodes transmit data packets to the
forwarder nodes and ND (i, s) is the latency occurred due to
direct transmission. Constraints in Eq. 36 and Eq. 37 indicate
that the ND and PD will be less than a minimum threshold for
successful transmission. If the delay increases then the packet
drop ratio increases.
1) GRAPHICAL ANALYSIS
Consider the scenario, where PD = 30 ms and ND = 20 ms
according to Eq. 38 and Eq. 39 which is represented as:
0 ≤ ND ≤ 20, (38)
20 ≤ PD ≤ 30, (39)
20 ≤ ND + PD ≤ 30. (40)
The intersection of corner points provided in Eq. 38 to Eq.
39 construct the solution region at upper right half plane, as
shown in Fig. 6. Each test point in the feasible region validates
the solution as:
at P1 (20, 0) = 20 ms,
at P2 (20, 30) = 30 ms,
and at P3 (0, 30) = 30 ms.
Therefore, all the intersecting points validate the feasible
region according to the given set of constraints.
6143
 N. Jan et al.: BECHA Algorithm for WSNs

FIGURE 7. Throughput: Feasible region.

FIGURE 6. End to end delay: Feasible region.

TABLE 2. Simulation parameters.

E. THROUGHPUT
During the transmission phase, each node sends data to either
sink directly or advances to the next hop forwarder nodes.
Throughput is defined as, the number of packets successfully
received at sink per unit of time. The data packets transmitted
by the sensor nodes is counted as one successful transmission,
while the data transmitted by the forwarder nodes as compos-
ite packet is also counted as one successful transmission. The
objective function is defined as:
rmax X
X n
Max Nprx ∀r ∈ R ∧ x ∈ N , (41)
r=1 x=1
such that, 1) GRAPHICAL ANALYSIS
Suppose a scenario with total bandwidth B between
C1 : Ex ≤ Exini ∀x ∈ N , (42) 2273 KHz to 3000 KHz. The B allocated to BF and BNF
C2 : Ex ≤ Ex−1 ∀x ∈ N , (43) according to Eq. 46, and Eq. 47 respectively:
C3 : dx ≤ dxmin ∀x ∈ N , (44)
fmax
rmax X 273 ≤ BF ≤ 1000, (46)
X
C4 : BF rf ∀r ∈ R ∧ f ∈ F. (45) 2000 ≤ BNF ≤ 2750, (47)
r=1 f =1 2073 ≤ BF + BNF ≤ 3000, (48)
Where, Eq. 41 is the objective function followed by con-
straints to maximize the successful packet delivery Np in The corner points considered in Eq. 46 to Eq. 48 generate the
each round. Constraint in Eq. 42 shows the energy bound i.e., feasible region on upper right half plane at intersecting points
nodes are attached with limited energy. Therefore, energy is P1 , P2 , and P3 respectively:
the important constraint in the entire transmission process. at P1 (250, 2000) = 2060 KHz,
Eq. 43 deals with the energy of outer and inner coronas nodes. at P2 (1000, 2000) = 3000 KHz,
Before the data transmission each node checks the energy at P3 (250, 2750) = 3000 KHz.
of its forwarder nodes to avoid the load dis-proportionality. Therefore, it is proved that all the points mentioned in Fig. 7
Whereas, constraint in Eq. 44 presents a distance metric validate the solution under given set of constraints.
threshold for successful communication. The aim of the
induced Eq. 45 is to minimize the load on those nodes which VI. PERFORMANCE EVALUATION
participates more in the data transmission and have low We perform simulations of our proposed protocols and com-
residual energy level. Alternatively, the load is reallocated pare with existing WSNEHA and ECMSE algorithms. The
to proximal neighbors with high residual energy. Where, BF evaluation is measured in term of throughput, number of dead
shows the bandwidth allocated to the forwarder nodes with nodes, energy consumption, number of packets drop, data
high energy level, and BNF is the corresponding bandwidth load on each corona nodes and end-to-end delay. The main
allocated to non-forwarder nodes. parameter setting is summarized in Table 2.

6144 VOLUME 5, 2017

N. Jan et al.: BECHA Algorithm for WSNs

FIGURE 8. The data size of sensor nodes increases with the decrease in FIGURE 9. Energy consumption of sensor nodes with varying distance
transmission range and distance. and transmission range.

TABLE 3. Transmission range with energy consumption and data size

Sensor nodes except the outer most corona nodes receive transmitted.
cumulative load. The internal corona nodes forward the
aggregated data packets as a composite packet to either sink
directly or to the next hop forwarding nodes. In order to for-
ward the aggregated data to sink, the sink proximal neighbors
are selected as relays to sink. Consequently, the nodes die
very early and causes energy hole problem. Despite the fact
that WSNEHA algorithm balances the load between the sink
proximal nodes, though energy hole problem emerges in the
onward corona. Similarly, the network of ECMSE disjoints
very early due to the iterative selection of nodes located
near the optimal position. To extend the work of WSNEHA
algorithm, BECHA is presented. Thereafter, EA-BECHA is
proposed to overcome the limitations of ECMSE and BECHA
algorithms.
A. DATA SIZE
Fig. 8 shows the data transmitted by sensor nodes with vary-
ing distance from the sink and with different transmission
ranges r. In multi-hop communication manner if r is smaller
then the data is mainly distributed in the first radius range
of the sink. The sensor nodes located in this corona commu-
nicate directly to the sink and responsible for the transmis- FIGURE 10. Accumulative number of data packets delivered to sink.
sion of aggregated data packets. Moreover, there exist some
maximal energy consuming nodes which participate more in
data transmission due to their closeness to the sink. If r is sensor nodes at smaller distance will participate more in the
increased then the second corona nodes mainly participate data transmission as depicted in Fig. 8. Consequently, the
in the load distribution. Similarly, by increasing r the packet energy exhausts very early due to their closeness. Similarly,
load decreases on the inner corona nodes and increases on the if r increases the energy consumption become decreases. For
outer corona nodes. The reason is that the sensor nodes act as instance, if r = 10, 0.9 × 104 bits of data is transmitted with
relay to sink. Consequently, the energy of nodes exhaust more an energy dissipation of 6 mJ . Similarly, when r is maximum
quickly and the energy hole problem is emerged. For instance, i.e., r = 100, 0.1 × 104 bits of data is transmitted with energy
if r = 10, 0.9×104 bits of data is transmitted to the sink by the consumption of 1 mJ as given in Table 3.
first corona nodes. Similarly, when r is maximum very low
C. THROUGHPUT
amount of data is transmitted to the sink. i.e., 0.04 × 104 bits.
Fig. 10 exhibits the higher throughput of EA-BECHA than
B. ENERGY CONSUMPTION that of BECHA, WSNEHA, and ECMSE. At the initial
Fig. 9 shows the energy consumption of sensor nodes with stages all the algorithms have same packet delivery ratio;
varying r and distance from the sink. It is observed that the however, the difference arises after 100 rounds. The

VOLUME 5, 2017 6145


EA-BECHA algorithm outperforms because each time an
optimal forwarder is selected to advance the data packet based
on high FF . As a result, more data is delivered to the sink
i.e., 3.5 × 104 pkts. Hence, increases the throughput 28%
more than WSNEHA and 99% more than ECMSE. On the
other hand BECHA performs better than the existing schemes
because of its good load balancing ability. The WSNEHA has
low throughput due to the dis-junction of network caused by
the maximal energy consuming nodes located in the 2nd and
onward coronas nodes. Hence, no more data is forwarded to
the sink after the emergence of energy hole. The ECMSE
performs worse because each time the forwarder nodes are
selected iteratively near the energy optimal position, which
results in the early exhaustion of energy and creates energy
hole problem.
FIGURE 11. Number of dead nodes in each round.
D. NETWORK LIFETIME AND NUMBER OF DEAD NODES
The number of dead nodes in each round is shown in Fig.
11. It is clear from the figure that our proposed schemes
outperform the ECMSE, and WSNEHA with considerable
margins because of the balanced energy consumption and
optimal forwarder selection strategy. The dead nodes ratio
of ECMSE increases very sharply because of imbalance
load distribution and also due to the selection of longer
routing path. Consequently, more energy is exhausted in
the transmission phase i.e., the first dead node emerges in
50th round, and all nodes die in 500 rounds. Similarly, due
to the energy hole problem caused by the maximal energy
consuming nodes in the 2nd corona and other region nodes,
WSNEHA performs low as compared to the proposed algo-
rithms. EA-BECHA further decreases the number of dead
nodes as depicted in Fig. 11. It can be seen that the first dead
node of WSNEHA appears in the 102 rounds because load of
only first corona nodes is balanced; however, the sensor nodes
of second and onward coronas still facing the load imbalance
issue.
The BECHA algorithm shows improvement in the network
lifetime longevity i.e., 13% and 110% more than WSNEHA
and ECMSE. Moreover, the first dead node occurs in
6146
N. Jan et al.: BECHA Algorithm for WSNs
approximately 2×102 rounds. While the EA-BECHA outper-
forms in prolonging the network lifetime i.e., 58% more than
WSNEHA and 166% high than the ECMSE. It is because the
algorithm selects best forwarder for data transmission which
is being energy efficient and energy balanced.
FIGURE 12. Energy consumption of sensor nodes located at various
distances from the sink.
E. ENERGY UTILIZATION
Fig. 12 shows the energy consumption of sensor nodes with
varying distance from the sink. The ECMSE consumes more
energy in the initial stage because of iterative selection of
sensor nodes. As it is clear from the figure that the energy
of sensors in the first corona is balanced by WSNEHA algo-
rithm. However, the remaining nodes of other coronas suffer
the problem of imbalance load distribution which results in
the early exhaustion of energy. Consequently, the energy hole
problem is emerged and results in low network lifetime. The
BECHA algorithm efficiently utilizes the energy resource and
balances the load distribution of entire network. Whereas,
EA-BECHA results in good utilization of energy resource
i.e., 25% less than WSNEHA and 51% less than ECMSE.
Because in each round an optimal forwarder is selected from
the minimum energy consuming nodes which is responsible
for the transmission of aggregated data. The method, iterates
until the energy level of sensor nodes in each corona reaches
an equilibrium state.
F. PACKETS DROP
The packets drop ratio of EA-BECHA is minimum than that
of the ECMSE, BECHA, and WSNEHA algorithms i.e., in
500 rounds the packets drop ratio is 35% less than WSNEHA
and 21% less than that of ECMSE as depicted in Fig. 13. It is
because of low mortality rate of sensor nodes which results in
energy hole avoidance. Moreover, the selection of forwarder
nodes based on the optimal distance also resist the packets
drop ratio i.e., the packet is transmitted to the forwarder nodes
with low error rate. The ECMSE algorithm also performs
well mainly because of minimum mean square error rate.
VOLUME 5, 2017
N. Jan et al.: BECHA Algorithm for WSNs
FIGURE 13. Number of packets drop in each round.
In contrast to the EA-BECHA and ECMSE, the WSNEHA
and BECHA result in more packets drop because of erroneous
location. As a result the packets are transmitted by sender
nodes, while not received at the destination.
FIGURE 14. Accumulative number of transmission to deliver the data
packets.
G. NUMBER OF TRANSMISSIONS
The number of transmissions is mainly dependent on the
number of hops, and total number of re-transmissions.
Fig. 14 illustrates the number of transmissions that are
required to forward the data packets to the sink. Initially, all
the algorithms have approximately same number of transmis-
sion. However, it diverges after 100 rounds mainly because
of forwarders selection criteria. The WSNEHA shows less
number of transmissions because no more data is transmitted
when the network becomes disjoint. Contrary, to the dis-
cussed algorithm, BECHA depicts high number of transmis-
sions because each time the data packet is handed over to the
adjacent neighbors which is selected between the local maxi-
mum and local minimum threshold. The EA-BECHA shows
moderate number of transmissions because the higher priority
nodes suppress the transmission of low priority nodes. There-
VOLUME 5, 2017
fore, the algorithm indicates good improvement and gives
moderate number of transmission with high packet delivery
ratio. The ECMSE has more number of transmissions because
of two reasons 1) more number of re-transmissions, and 2)
due to longer routing path selection.
FIGURE 15. Average end-to-end delay of sensor nodes during data
delivery.
H. END TO END DELAY
The comparative analysis shows that the WSNEHA algo-
rithm has low latency rate i.e., 0.2 × 10−6 s as compared
to the ECMSE and proposed algorithms as illustrated in
Fig. 15. The ECMSE algorithm has high delay is mainly
because of more number of re-transmissions and error cop-
ing. Furthermore, the BECHA and EA-BECHA have high
delay because of the selection strategy. Each time an opti-
mal forwarder is selected for the delivery of data packets.
Consequently, the algorithms take more time for data deliv-
ery. In high traffic scenario the proposed algorithms per-
form inferior than that of WSNEHA. Moreover, it is also
clear from the figure that the propagation delay is low when
the distance between the nodes and sink is minimum, and
conversely.
I. LOAD DISTRIBUTION
Fig. 16 shows the data packets distributed to each corona
nodes. As depicted in figure that the data load is distributed
uniformly between the sink proximal neighbors. It is because
of the fact that each time the node with high residual energy
is selected as forwarder node. However, WSNEHA algorithm
shows dis-proportionality in data distribution among nodes in
onward coronas when r is low. The ECMSE, distributes the
data load unevenly between each corona nodes. As shown
in Fig. 16 that there is immense data load on inner coronas
nodes i.e., 1.6 × 103 pkts. Moreover, the unequal load is
distributed to the sensor nodes lie on the energy optimal posi-
tion. Contrary to the existing algorithms, the load balancing
is achieved by the proposed algorithms. It is because of the
good balancing ability and efficient forwarder nodes selection
strategy.
6147

TABLE 4. Performance trade-off of proposed algorithms with existing algorithms.
FIGURE 16. Data packet delivered to sensor nodes.
VII. PERFORMANCE TRADE-OFF
In this section the improvements and limitation of proposed
algorithms with existing algorithms are discussed. The com-
parative analysis of proposed with previous algorithms show
good improvement in energy consumption, network life-
time, throughput and packet drop ratio as given in Table 4.
We achieved network lifetime and throughput by balancing
the load distribution uniformly. Such as EA-BECHA pro-
longs the network lifetime with 58% more than WSNEHA
and 166% more than ECMSE with energy efficiency of
25% and 51%, respectively. Similarly, the throughput is
enhanced with 28% and 99% with less number of packet
drop ratio i.e., 35% and 21%. On the contrary, EA-BECHA
has higher end to end delay mainly because of processing
time for selection of forwarder nodes i.e., 110% higher than
WSNEHA and 90% high than that of ECMSE algorithm.
Similarly, BECHA has high end to end delay of 20% and 10%
than that of the existing algorithms.
VIII. CONCLUSION AND FUTURE WORK
Energy balancing and energy hole problem are the key
issues in maximizing the lifetime of a network. In WSNs
due to unbalanced load distribution sensor nodes die very
early which leads to energy hole problem. Moreover, due to
erroneous location information the data is not successfully
delivered to the neighbor nodes causing low throughput.
6148
N. Jan et al.: BECHA Algorithm for WSNs
To balance the energy consumption efficiently and alleviate
the energy hole problem, we propose BECHA. Thereafter,
EA-BECHA is presented to enhance the throughput and min-
imize the packets drop ratio. This work mainly adopts optimal
forwarding strategy to utilize the energy resource efficiently.
Moreover, our forwarder node selection mechanism reduces
the overall transmission distance. Thus, balances the load
distribution and achieves energy efficiency. Moreover, due to
low expected error rate the EA-BECHA achieves high packet
delivery ratio with less number of packets drop. Finally,
the simulation results indicate effectiveness of the proposed
algorithm which shows superiority in network lifetime and
throughput on the cost of increased end-to-end delay.
In future work we intend to apply this strategy in UWSNs.
The energy hole detection and alleviation would be more
challenging when the mobility of deployed nodes is consid-
ered in 3-D underwater environment. An adaptive mechanism
for the selection of transmission range and forwarding nodes
will be followed to avoid energy hole problem. Moreover, to
reduce the end-to-end delay an optimal setting for holding
time would also be considered. Furthermore, the incorpora-
tion of buffer-aided relays would be an interesting aspect to
explore.
REFERENCES
[1] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, ‘‘A survey
on sensor networks,’’ IEEE Commun. Mag., vol. 40, no. 8, pp. 102–114,
Aug. 2002.
[2] K. Akkaya and M. Younis, ‘‘A survey on routing protocols for wireless
sensor networks,’’ Ad Hoc Netw., vol. 3, no. 3, pp. 325–349, 2005.
[3] N. A. Pantazis, S. A. Nikolidakis, and D. D. Vergados, ‘‘Energy-efficient
routing protocols in wireless sensor networks: A survey,’’ IEEE Commun.
Surveys Tuts., vol. 15, no. 2, pp. 551–591, 2nd Quart., 2013.
[4] S. Zeadally, S. U. Khan, and N. Chilamkurti, ‘‘Energy-efficient net-
working: Past, present, and future,’’ J. Supercomput., vol. 62, no. 3,
pp. 1093–1118, 2012.
[5] M. Magno, D. Boyle, D. Brunelli, E. Popovici, and L. Benini, ‘‘Ensur-
ing survivability of resource-intensive sensor networks through ultra-low
power overlays,’’ IEEE Trans. Ind. Informat., vol. 10, no. 2, pp. 946–956,
May 2014.
[6] X. Wu, G. Chen, and S. K. Das, ‘‘Avoiding energy holes in wireless
sensor networks with nonuniform node distribution,’’ IEEE Trans. Parallel
Distrib. Syst., vol. 19, no. 5, pp. 710–720, May 2008.
[7] C.-F. Wang, J.-D. Shih, B.-H. Pan, and T.-Y. Wu, ‘‘A network lifetime
enhancement method for sink relocation and its analysis in wireless sensor
networks,’’ IEEE Sensors J., vol. 14, no. 6, pp. 1932–1943, Jun. 2014.
[8] W. Raza et al., ‘‘An improved forwarding of diverse events with mobile
sinks in underwater wireless sensor networks,’’ Sensors, vol. 16, no. 11,
pp. 1850–1875, 2016.
VOLUME 5, 2017
N. Jan et al.: BECHA Algorithm for WSNs
[9] S. Olariu and I. Stojmenovic, ‘‘Design guidelines for maximizing lifetime
and avoiding energy holes in sensor networks with uniform distribution
and uniform reporting,’’ in Proc. INFOCOM, Apr. 2006, pp. 1–12.
[10] M. A. Khan, A. Sher, A. R. Hameed, N. Jan, J. S. Abassi, and N. Javaid,
‘‘Network lifetime maximization via energy hole alleviation in wireless
sensor networks,’’ in Proc. Int. Conf. Broadband Wireless Comput. Com-
mun. Appl., 2016, pp. 279–290.
[11] Z.-W. Zeng, Z.-G. Chen, and A.-F. Liu, ‘‘Energy-hole avoidance for WSN
based on adjust transmission power,’’ Chin. J. Comput., vol. 33, no. 1,
pp. 12–22, 2010.
[12] X. Liu, ‘‘A novel transmission range adjustment strategy for energy hole
avoiding in wireless sensor networks,’’ J. Netw. Comput. Appl., vol. 67,
pp. 43–52, May 2016.
[13] Y. Xue, X. Chang, S. Zhong, and Y. Zhuang, ‘‘An efficient energy hole
alleviating algorithm for wireless sensor networks,’’ IEEE Trans. Consum.
Electron., vol. 60, no. 3, pp. 347–355, Aug. 2014.
[14] P. K. Sahoo and W.-C. Liao, ‘‘HORA: A distributed coverage hole repair
algorithm for wireless sensor networks,’’ IEEE Trans. Mobile Comput.,
vol. 14, no. 7, pp. 1397–1410, Jul. 2015.
[15] K. Latif, N. Javaid, A. Ahmad, Z. A. Khan, N. Alrajeh, and M. I. Khan, ‘‘On
energy hole and coverage hole avoidance in underwater wireless sensor
networks,’’ IEEE Sensors J., vol. 16, no. 11, pp. 4431–4442, Jun. 2016.
[16] X. Liu, ‘‘An optimal-distance-based transmission strategy for lifetime
maximization of wireless sensor networks,’’ IEEE Sensors J., vol. 15, no. 6,
pp. 3484–3491, Jun. 2015.
[17] A. M. Popescu, N. Salman, and A. H. Kemp, ‘‘Energy efficient geographic
routing robust against location errors,’’ IEEE Sensors J., vol. 14, no. 6,
pp. 1944–1951, Jun. 2014.
[18] N. Jan, A. R. Hameed, B. Ali, R. Ullah, K. Ullah, and N. Javaid, ‘‘A
wireless sensor networks balanced energy consumption protocol,’’ in Proc.
31st IEEE Int. Conf. Adv. Inf. Netw. Appl. Workshops (WAINA), Taipei,
Taiwan, 2017.
[19] Y.-S. Chen and Y.-W. Lin, ‘‘Mobicast routing protocol for underwater
sensor networks,’’ IEEE Sensors J., vol. 13, no. 2, pp. 737–749, Feb. 2013.
[20] R. W. L. Coutinho, A. Boukerche, L. F. M. Vieira, and A. A. F. Loureiro,
‘‘Geographic and opportunistic routing for underwater sensor networks,’’
IEEE Trans. Comput., vol. 65, no. 2, pp. 548–561, Feb. 2016.
[21] C. Zidi, F. Bouabdallah, and R. Boutaba, ‘‘Routing design avoiding energy
holes in underwater acoustic sensor networks,’’ Wireless Commun. Mobile
Comput., vol. 16, no. 14, pp. 2035–2051, Oct. 2016.
[22] M. Abo-zahhad, S. M. Ahmed, N. Sabor, and S. Sasaki, ‘‘Mobile sink-
based adaptive immune energy-efficient clustering protocol for improv-
ing the lifetime and stability period of wireless sensor networks,’’ IEEE
Sensors J., vol. 15, no. 8, pp. 4576–4586, Aug. 2015.
[23] C. Tunca, S. Isik, M. Y. Donmez, and C. Ersoy, ‘‘Ring routing:
An energy-efficient routing protocol for wireless sensor networks
with a mobile sink,’’ IEEE Trans. Mobile Comput., vol. 14, no. 9,
pp. 1947–1960, Sep. 2015.
[24] H. Salarian, K.-W. Chin, and F. Naghdy, ‘‘An energy-efficient mobile-sink
path selection strategy for wireless sensor networks,’’ IEEE Trans. Veh.
Technol., vol. 63, no. 5, pp. 2407–2419, Jun. 2014.
[25] K. Lin, M. Chen, S. Zeadally, and J. J. P. C. Rodrigues, ‘‘Balanc-
ing energy consumption with mobile agents in wireless sensor net-
works,’’ Future Generat. Comput. Syst., vol. 28, no. 2, pp. 446–456,
2012.
[26] S. Saleh, M. Ahmed, B. M. Ali, M. F. A. Rasid, and A. Ismail, ‘‘A self-
optimizing scheme for energy balanced routing in wireless sensor networks
using sensorant,’’ Sensors, vol. 12, no. 8, pp. 11307–11333, 2012.
[27] Y. Liao, H. Qi, and W. Li, ‘‘Load-balanced clustering algorithm with dis-
tributed self-organization for wireless sensor networks,’’ IEEE Sensors J.,
vol. 13, no. 5, pp. 1498–1506, May 2013.
[28] Z. Han, J. Wu, J. Zhang, L. Liu, and K. Tian, ‘‘A general self-organized
tree-based energy-balance routing protocol for wireless sensor network,’’
IEEE Trans. Nucl. Sci., vol. 61, no. 2, pp. 732–740, Apr. 2014.
[29] J. Li and P. Mohapatra, ‘‘Analytical modeling and mitigation techniques for
the energy hole problem in sensor networks,’’ Pervasive Mobile Comput. J.,
vol. 3, no. 3, pp. 233–254, Jun. 2007.
[30] J. Ren, Y. Zhang, K. Zhang, A. Liu, J. Chen, and X. S. Shen, ‘‘Lifetime
and energy hole evolution analysis in data-gathering wireless sensor net-
works,’’ IEEE Trans. Ind. Informat., vol. 12, no. 2, pp. 788–800, Apr. 2016.
[31] L. Kong, K. Ma, B. Qiao, and X. Guo, ‘‘Adaptive relay chain routing
with load balancing and high energy efficiency,’’ IEEE Sensors J., vol. 16,
no. 14, pp. 5826–5836, Jul. 2016.
VOLUME 5, 2017
[32] M. A. Abd, S. F. Majed Al-Rubeaai, B. K. Singh, K. E. Tepe, and
R. Benlamri, ‘‘Extending wireless sensor network lifetime with global
energy balance,’’ IEEE Sensors J., vol. 15, no. 9, pp. 5053–5063, Sep. 2015.
[33] K.-V. Nguyen, P. L. Nguyen, Q. H. Vu, and T. Van Do, ‘‘An energy efficient
and load balanced distributed routing scheme for wireless sensor networks
with holes,’’ J. Syst. Softw., vol. 123, pp. 92–105, Jan. 2017.
[34] C. Sha, H. Chen, C. Yao, Y. Liu, and R.-C. Wang, ‘‘A type of energy hole
avoiding method based on synchronization of nodes in adjacent annuluses
for sensor network,’’ Int. J. Distrib. Sensor Netw., vol. 12, no. 3, pp. 1–14,
2016.
[35] Z. Xu, L. Chen, T. Liu, L. Cao, and C. Chen, ‘‘Balancing energy con-
sumption with hybrid clustering and routing strategy in wireless sensor
networks,’’ Sensors, vol. 15, no. 10, pp. 26583–26605, 2015.
[36] H. M. Ammari, ‘‘Investigating the energy sink-hole problem in connected
k-covered wireless sensor networks,’’ IEEE Trans. Comput., vol. 63,
no. 11, pp. 2729–2742, Nov. 2014.
[37] Y. Zhang, X. Zhang, W. Fu, Z. Wang, and H. Liu, ‘‘HDRE: Coverage hole
detection with residual energy in wireless sensor networks,’’ J. Commun.
Netw., vol. 16, no. 5, pp. 493–501, Oct. 2014.
[38] G. Zhang, X. Wu, J. Chen, X. Wang, and X. Yan, ‘‘On the problem of
energy balanced relay sensor placement in wireless sensor networks,’’ Int.
J. Distrib. Sensor Netw., vol. 9, no. 10, pp. 1–9, 2013.
[39] A. Ahmad, N. Javaid, Z. A. Khan, U. Qasim, and T. A. Alghamdi,
‘‘(ACH)2 : Routing scheme to maximize lifetime and throughput of wire-
less sensor networks,’’ IEEE Sensors J., vol. 14, no. 10, pp. 3516–3532,
Oct. 2014.
NAEEM JAN received the B.S. degree in com-
puter science from PMAS, University Institute
of Information Technology Rawalpindi, Pakistan,
in 2014, and the M.S. degree in computer sci-
ence, under the supervision of Dr. N. Javaid, from
the Department of Computer Science, COMSATS
Institute of Information Technology (COMSATS
IIT), Islamabad, Pakistan, in 2017. He is cur-
rently with ComSens (Communication over Sen-
sors) Research Laboratory, COMSATS IIT. His
research interests include wireless sensor networks, optimization techniques,
and Internet of Things.
NADEEM JAVAID (S’8–M’11–SM’16) received
the master’s degree in electronics from Quid-I-
Azam University, Islamabad, Pakistan, and the
Ph.D. degree from the University of Paris-Est,
France, in 2010. He is currently an Associate
Professor and the Founding Head of the Com-
Sens (Communications over Sensors) Research
Group, Department of Computer Science, COM-
SATS Institute of Information Technology, Islam-
abad, Pakistan. He has supervised more than 70
M.S. and six Ph.D. theses. He has authored more than 400 papers in peer-
reviewed technical journals and international conferences. His research inter-
ests include wireless ad hoc networks, wireless sensor networks, and energy
optimization in smart grids.
QAISAR JAVAID is currently an Assistant Pro-
fessor with the Department of Computer Science
Software Engineering, International Islamic Uni-
versity Islamabad. He is also heading the CISCO
Networking Academy of International Islamic
University, Islamabad, Pakistan. His research
interests include computer networks, information
security, cloud computing, and Internet of Things.
6149

NABIL ALRAJEH received the Ph.D. degree in
biomedical informatics engineering from Vander-
bilt University, USA. He was a Senior Advi-
sor for the Ministry of Higher Education, where
he implemented development programs, including
educational affairs, health information systems,
strategic planning and research, and innovation.
He is currently an Associate Professor of Medi-
cal Informatics with the Biomedical Technology
Department, King Saud University. His research
interests include e-Health applications, hospital information systems,
telemedicine, intelligent tutoring systems, and wireless sensor networks.
MASOOM ALAM received the Ph.D. degree in
computer science from the University of Inns-
bruck, Austria, in 2007.
He was a Lecturer in IM sciences. He has 10
years’ experience in teaching, system administra-
tion, and research and development. He is leading
a research group with more than 35 graduates and
undergraduates.
6150
N. Jan et al.: BECHA Algorithm for WSNs
ZAHOOR ALI KHAN (SM’15) received the B.Sc.
degree from the University of Peshawar, the M.Sc.
degree in computer engineering from UET Tex-
ila, the M.Sc. degree in electronics from Quaid-
i-Azam University, Islamabad, Pakistan, the Ph.D.
degree from the Faculty of Engineering, Dalhousie
University, Canada, and the M.C.Sc degree from
the Faculty of Computer Science, Dalhousie Uni-
versity. He was an Assistant Professor (PT) with
the Department of Engineering Mathematics and
Internetworking, a Post-Doctoral Fellow (Internetworking Program) with the
Faculty of Engineering, Dalhousie University, and a part-time Professor of
Computing and Information Systems with Saint Mary’s University, Nova
Scotia, Canada. He is currently an Assistant Professor with CIS, Higher
Colleges of Technology, Fujairah Campus, UAE. He has more than 15 years
of research and development, academia, and project management experience
in the IT and engineering fields. He has authored or co-authored a book and
more than 180 peer-reviewed journal and conference papers. He has multi-
disciplinary research skills on emerging wireless technologies. His research
interests include, but are not limited to, the areas of e-Health pervasive
wireless applications, theoretical and practical applications of wireless (Body
Area) sensor networks, and software-defined networks. He is interested in
designing and implementing the algorithms related to energy and quality-
of-service-aware routing protocols, fault management, security, and privacy.
He serves as a Regular Reviewer/Organizer of numerous reputed ISI-indexed
journals, IEEE conferences, and workshops. He is a Senior Member of the
IEEE Communication Society and the IAENG.
IFTIKHAR AZIM NIAZ is currently a Profes-
sional Engineer and an Assistant Professor, and
the Ph.D. Coordinator with the Department of
Computer Science. He has been coordinating with
different government and private sectors and orga-
nizations, improving the academic standard of stu-
dents. He has been the Ph.D. Supervisor of the
HEC since 2007. Because of his professional expe-
rience, expertise, social and communication skills,
he has been on the expert panel of the PEC Accred-
itation Team since 2009. He is also the member of various national and
international professional bodies and clubs that is one of the necessary
conditions for managing projects. Since 2007, he has been a Joint Secretary
of the MEXT Alumni Association of Pakistan.
VOLUME 5, 2017